INDEX

Explanations

relevant content related to full context and additional information within a given document

oai_token-act-pair · gpt-3.5-turbo

mentions of "full" or "fullest" in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Kimmel

-0.74

 peripher

-0.69

 Tackle

-0.66

ovan

-0.65

Tup

-0.65

aughs

-0.64

 Shack

-0.63

*/(

-0.62

 Beir

-0.61

 Michaels

-0.60

POSITIVE LOGITS

withstanding

0.88

ature

0.87

theless

0.84

ening

0.82

sheet

0.81

full

0.81

iful

0.80

ifully

0.80

":""},{"

0.80

ntil

0.79

Activations Density 0.004%

relevant content related to full context and additional information within a given document

mentions of "full" or "fullest" in various contexts

No Comments

No Known Activations